key: cord-0292505-gbnkuw1c
authors: Semple, SL; Jacob, RA; Mossman, K; DeWitte-Orr, SJ
title: Long dsRNA mediated RNA interference (dsRNAi) is antiviral in interferon competent mammalian cells
date: 2022-01-20
journal: bioRxiv
DOI: 10.1101/2022.01.14.476298
sha: 55ffb3d398bd44d4f88ad414223fbe815afdf211
doc_id: 292505
cord_uid: gbnkuw1c

In invertebrate cells, RNA interference (RNAi) acts as a powerful defense against virus infection by cleaving virally produced long dsRNA into siRNA by Dicer and loaded into RISC which can then destroy/disrupt complementary viral mRNA sequences. Comparatively in mammalian cells, the type I interferon (IFN) pathway is the cornerstone of the innate antiviral response. Although the cellular machinery for RNAi functions in mammalian cells, its role in the antiviral response remains controversial. Here we show that IFN competent mammalian cells engage in dsRNA-mediated RNAi. We found that pre-soaking mammalian cells with concentrations of sequence-specific dsRNA too low to induce IFN production could significantly inhibit viral replication, including SARS-CoV-2. This phenomenon was dependent on dsRNA length, was comparable in effect to transfected siRNAs, and could knockdown multiple sequences at once. Additionally, Dicer-knockout cell lines were incapable of this inhibition, confirming use of RNAi. This represents the first evidence that soaking with gene-specific dsRNA can generate viral knockdown in mammalian cells. Furthermore, demonstrating RNAi below the threshold of IFN induction has uses as a novel therapeutic platform.

cells have normal IFN function and are exposed to long dsRNA, it has been shown that the IFN pathway 151 actively inhibits RNAi (Seo et al., 2013 ; Van der Veen et al., 2018) . However, none of these studies soaked 152 cells with long dsRNA at concentrations that were too low to induce the IFN response. Moreover, many of 153 these studies either transfect cells with long dsRNA or use the TLR3 agonist, polyinosinic:polycytidylic 154 acid (pIC, reviewed by Komal et al., 2021) . The use of pIC is an excellent tool for understanding the IFN 155

pathway, but it is important to note that this molecule is not the same as naturally occurring dsRNA. It has 156 no defined length, a preparation of pIC can range from 1.5 kb to 8 kb and contains complimentary strands 157 of inosines and cytosines, that would be not found in nature, to produce a dsRNA helix (Scadden, 2007) . 158

Thus, the results from both pIC and dsRNA transfection studies may not be indicative of the natural cellular 159 responses to extracellular dsRNA, particularly at low concentrations. This suggests a fascinating possibility, 160

where RNAi is the mechanism used by mammalian cells when dsRNA levels are too low to induce IFN. 161

This sentinel activity could provide pre-emptive protection and/or clearance early in the course of infection 162 when viral numbers are not yet high enough to warrant the costly use of the IFN pathway. 163

Since its discovery in 1998 by Fire and colleagues, scientists have been fascinated with the gene 164 knockdown potential of RNAi. Yet, as described above, this sequence-specific knockdown did not seem 165 possible in IFN-competent mammalian cells without the use of transfection agents. Moreover, the 166 understanding of how cells respond to non-IFN inducing concentrations of dsRNA is completely absent 167 from the literature. In the present study, we provide evidence that antiviral RNAi can be induced in 168 mammalian cells by simply pre-soaking the cells with dsRNA at concentrations that are too low to induce 169 IFN production. Remarkably, we were able to demonstrate this phenomenon in multiple mammalian cell 170 types using several different dsRNA sequences to inhibit the infection of vesicular stomatitis virus 171 expressing green fluorescent protein (VSV-GFP), as well as the human coronaviruses (CoV) HCoV-229E 172

and SARS-CoV-2. Additionally, we reveal that this phenomenon is length-dependent and requires the 173 presence of Dicer. Aside from the implications this work could have on developing novel antiviral/gene 174 therapies, these results provide an explanation as to why the mammalian lineage retained all the necessary 175 machinery for RNAi and why several mammalian viruses have devoted parts of their valuable genetic 176 material to inhibit this pathway (Wang et al., 2006; Yang et al., 2013; Qui et al., 2020) . 177 178 2. Results: 179

3

Because RNAi appears to be masked by the interferon response, it was crucial to identify which 181 concentration of soaked dsRNA would not induce the IFN pathway. When gene expression of IFNβ and 182 CXCL10 was measured 26h after cells were exposed to 700 bp GFP dsRNA (0.5 μg/mL and 10 μg/mL) or 183

10 μg/mL of HMW pIC, only the pIC condition appeared to induce the IFN response (Figure 1) . The THF 184

and SNB75 cell lines were initially selected for this study to explore whether both a normalized cell line 185 (THF) and an "abnormal" cancerous cell line (SNB75) would be capable of long dsRNAi while being IFN 186 competent. In THF, the gene expression of IFNβ increased only in the pIC exposure condition, but due to 187 variability this was not significantly different from the other conditions (Figure 1Ai) . When SNB75 was 188 stimulated with these dsRNA and pIC doses, only the pIC treatment was able to induce significant 189 upregulation of IFNβ gene expression (Figure 1Bi) . Because IFNβ gene expression is known to be quite 190 rapid and short-lived, the more persistent ISG, CXCL10, was also measured. In both THF and SNB75, 191 CXCL10 gene expression was only observed to significantly increase when cells were soaked with pIC 192 (Figures 1Aii and 1Bii) . Neither dsRNA concentration appeared to induce significant upregulation of IFNβ 193 or CXCL10 when compared to the unstimulated control (Figure 1) . When comparing the molar amounts, 194

5.1 nM of pIC (average length of 3000 bp) and 21.6 nM of 700 bp dsRNA was added to the cells. This 195 means that four times more dsRNA molecules were added to each cell when compared to the number of 196 pIC molecules. Furthermore, pIC is over four times longer than the dsRNA molecules used here, so it is 197 difficult to compare efficacy of IFN induction between these molecules. As such, pIC should only be 198 considered a positive control in this experiment. 199

3.2 Soaking with long dsRNA does not negatively impact the viability of mammalian cells 200

To validate that soaking with long dsRNA does not negatively influence the health status of THF, 201

SNB75 or MRC5, cell survival and metabolism were both measured. Following 24h exposure to a range of 202 700 bp GFP dsRNA concentrations, cellular metabolism was shown to significantly increase at only the 203 highest dsRNA concentration assessed, 800 ng/mL (Figure 2A) . The toxicity experiments did not use 204 dsRNA concentrations greater than 800 ng/mL because higher concentrations were unnecessary to see 205

RNAi effects. The significant increase in cellular metabolism at 800 ng/mL was observed in THF ( Figure  206 2Ai), SNB75 (Figure 2Aii ) and MRC5 (Figure 2Aiii ) when compared to the 0 ng/mL control. Meanwhile, 207 membrane integrity was shown to not be influenced at any of the dsRNA concentrations assessed in all 208 three of the cell lines studied ( Figure 2B) . None of the dsRNA treated cells presented values significantly 209

lower than the control cells, for both Alamar Blue and CFDA, indicating none of the dsRNAs treated were 210 cytotoxic. 211

3.3 Long dsRNAi can only be stimulated by dsRNA lengths of 400 bp or greater 212

It was initially observed that pre-soaking cells with 500 ng/mL of 700 bp GFP dsRNA for 2h could 213 stimulate protection towards VSV-GFP in both THF and SNB75 (Figure 3) . Cell viability (Figure 2 ) and 214

IFN induction by dsRNA (Figure 1 ) were both measured using 700bp long GFP dsRNA; however, the 215 length of dsRNA capable of inducing dsRNAi required optimization. It was observed that dsRNA of 300 216 bp and shorter could not significantly induce knockdown of VSV-GFP in THF cells ( Figure 3A ). In the 217 cancerous SNB75 cell line, the dsRNA length cut-off was less definitive as both 300 and 400 bp did not 218 significantly differ from either the control condition or those inducing significant knockdown ( Figure 3B ).

The appearance of the VSV-GFP infected THF following the dsRNA treatments revealed whether 220 knockdown was occurring as the level of fluorescence is directly related to viral load ( Figure 3C ). 221

Because GFP is not a naturally occurring gene found in viruses, the ability of dsRNA encoding viral 223 gene sequences to stimulate dsRNAi was explored next. Soaking mammalian cells with viral gene specific 224 dsRNA was shown to induce knockdown of corresponding viruses (Figure 4) . When THF and SNB75 were 225

pre-soaked with 500 ng/mL of 700 bp dsRNA synthesized to the N and M protein genes of VSV, significant 226 knockdown was observed when compared to the non-sequence matched controls of mCherry and Beta-lac 227

( Figure 4A and 4B) . Additionally, when a mixture of 250 ng/mL N protein dsRNA and 250 ng/mL M 228 protein dsRNA was used to pre-soak THF cells, significant knockdown was still observed but was 229 comparable to when only 500 ng/mL of either dsRNA was used ( Figure 4A ). For SNB75, this mixture pre-230 exposure was not observed to be significantly different to the control ( Figure 4B ). When MRC5 cells were 231

pre-soaked with 500 ng/mL of 700 bp dsRNA synthesized to the RdRp, N protein, M protein and spike 232 protein genes of HCoV-229E, significant reduction of viral particle production was observed for all 233 exposures except for the RdRp dsRNA ( Figure 4C ). In Calu-3 cells, significant reduction in viral 234

replication was observed after pre-treatment with 1000 ng/mL of SARS-CoV-2 N protein dsRNA when 235 compared to both the virus alone control and the mis-matched mCherry dsRNA control ( Figure 4D ).

However, no significant viral inhibition was observed when Calu-3 cells were pre-treated with 1000 ng/mL 237 of SARS-CoV-2 M protein dsRNA ( Figure 4D ). 238

3.5 Knockdown via dsRNA soaking is also observed in human pBECs 239

In addition to the immortalized cell lines described above, the knockdown capability of pre-soaking 240 cells with long dsRNA was also explored in primary pBEC cultures ( Figure 5 ). An image of the pBECs 241 after growth in culture for 28 days ( Figure 5A ). Significant knockdown of VSV-GFP was observed when 242 pBECs were pre-treated with 500 ng/mL of dsRNA to the N protein of the virus when compared to the 243 unmatched mCherry control ( Figure 5B) . Similar viral knockdown was also observed when the pBECs 244

were pre-soaked with HCoV-229E M protein dsRNA which resulted in significant knockdown of HCoV-245

229E when compared to the mCherry control ( Figure 5C ). As a comparison it was also shown that soaking 246 pBECs with 50 μg/mL of pIC also induced antiviral protection ( Figure 5C ). Indeed the level of protection 247

provided by pIC was comparable to that provided by M protein encoding dsRNA (p = 0.1053442). 248

3.6 Soaking cells with siRNA did not induce viral knockdown 249

When both THF and SNB75 cells were pre-soaked with GFP siRNA, TCID50 levels of VSV-GFP were 250

comparable to the unstimulated control and to the mis-matched long dsRNA mCherry control (Figure 6Ai  251 and 6Bi). Meanwhile, soaking these cells with 700 bp GFP dsRNA was again shown to induce significant 252 knockdown of the VSV-GFP virus (Figure 6Ai and 6Bi) . This result was not due to inefficacy of the siRNA 253 molecules as transfecting THF and SNB75 with the GFP siRNA induced significant knockdown when 254 compared to transfection with the negative control siRNA (Figure 6Aii and 6Bii ). At the timepoint tested 255 the knockdown of VSV-GFP by GFP siRNA is similar to that by 700 bp GFP dsRNA. 256 3.7 Long, synthetic combination dsRNA molecules can inhibit VSV-GFP via multiple gene knockdown 257

Combination dsRNA molecules were synthesized to test whether multiple VSV genes could be knocked 258

down when 700 bp of dsRNA contained sequences for two different viral genes. Figure 7A is a schematic 259 of the three different combination dsRNA molecules that were synthesized for this study. When THF cells 260

were pre-treated with 500 ng/mL of each combination dsRNA molecule and the mCherry unmatched 261 sequence control, only the three combination molecules were able to induce significant knockdown of VSV-262 GFP (Figure 7Bi ). When measuring gene expression, only the 5'N-3'M molecule was able to induce 263 significant knockdown of both the VSV N protein and M protein genes (Figure 7Bii and 7Biii) . When THF 264 cells were pre-exposed to 1000 ng/mL of the combination dsRNA molecules and the mCherry control, it 265

was observed again that only the three combination molecules induced significant knockdown of VSV- Knockdown of VSV-GFP was also obtained in the mouse MSC cell line that contains functional Dicer1 271

when pre-soaked with long dsRNA containing N protein sequence for 2h prior to infection ( Figure 8A ). In 272 comparison, when using the matching cell line that was Dicer1-defective, the significant decrease in viral 273 knockdown was abolished ( Figure 8B) murine cells with low concentrations of pIC (0.1-3 μg/mL) did not stimulate protein production of IFNβ 281 and CXCL10. However, when the cells were transfected with these low doses of pIC, a significant increase 282 was observed for both IFNβ and CXCL10 protein production (Hägele et al., 2009 ). In the present study, we 283 soaked human cells with 10 μg/mL of pIC and were successfully able to stimulate the gene expression of 284

IFNβ and the ISG, CXCL10. Furthermore, we observed that the long dsRNA concentrations used in the 285 present study did not induce the expression of these genes in both THF and SNB75. For the fibroblast cells 286 used, IFNβ would be anticipated to be the primary IFN produced in response to dsRNA (Li et al., 2018 soaking the cells with luciferase dsRNA that was not observed when transfecting them (Saleh et al., 2006) . 320

Though significant knockdown was still observed when soaking with shorter lengths, 200 bp and greater 321

were found to be much more effective at inducing luciferase knockdown (Saleh et al., 2006) . Because there 322

is no concern of stimulating the IFN response in invertebrate cells, concentration may also play a role that 323 cannot be explored in IFN-competent mammalian cells. This may provide an explanation as to why the 324 length requirement (~300-400 bp and greater) observed in the present study was greater than those 325 described using invertebrate cells. It is also possible that the size specificity of SID-1 includes smaller 326 dsRNA molecules when compared to SIDT2. Additionally, because the SIDT2 channel has a higher binding 327 affinity for dsRNA lengths ranging from 300-700 bp (Li et al., 2015) , this also supports its involvement 328

here wherein knockdown was only achievable in THF and SNB75 using dsRNA lengths of 300-400 bp and 329

greater.

Inhibition those that were very effective in the MDCK cells had protective effects in vivo (Ge et al., 2003) . When 339 exploring the use of siRNA for combatting COVID-19, Wu and Luo (2021) reported 50% inhibition rates 340 in 24h when targeting the structural Spike, N and M protein genes of SARS-CoV-2 that were overexpressed 341 in human epithelial cells. These results have also been replicated in live animal trials. In an in vivo trial, 342 mice were injected with lentiviruses containing siRNA that targeted either the L (polymerase) or N 343

(nucleocapsid) protein of the rabies virus (RV). It was found that targeting the structural N protein provided 344 62% protection to RV infected mice while no protection was observed when the L protein was the target 345 (Singh et al., 2014) . Based on these previous results, it appears that the type of virus, and likely, variances 346

in replication processes, play a role in which target genes have higher efficiency for RNAi knockdown. 347

When exploring a rhabdovirus and two coronaviruses in the current study, pre-soaking with long dsRNA 348 matching structural genes (N, M and spike proteins) was observed to be more successful than those 349 associated with the viral transcriptional machinery (RdRp). A systematic study of each gene, including 350

sequences within each gene, is needed in future studies to better understand what sequences are optimal 351 targets for suppressing virus replication via dsRNAi.

As treatments that stimulate dsRNAi towards a single viral gene were successful, so simultaneous 353

inhibition of multiple viral genes would be anticipated to enhance this effect. Combination treatments with 354

siRNAs have shown promise in the suppression of various viral pathogens. When siRNAs that targeted 355

both the G (glycoprotein) and the N protein genes of rabies virus was expressed in mammalian cells using 356 a single cassette, an 87% reduction of the target virus was observed (Meshram et al., 2013) . It should be 357 noted that individual sequences offered an 85% reduction in virus titres. Similarly, when rat fibroblast cells 358

were exposed to combination siRNAs targeting both the Immediate-early-2 and DNA polymerase genes, a 359 significant reduction in associated mRNAs and cytopathic effects was observed following infection with a 360 novel rat Cytomegalovirus (Balakrishnan et al., 2020) . This siRNA combination inhibition has also been 361 explored in vivo using both rhesus monkeys and macaques. SiRNA combinations targeting multiple genes 362 of the Zaire Ebola virus (ZEBOV) provided 66% protection in the rhesus monkeys and 100% protection in 363 macaques to lethal doses of ZEBOV when this treatment was administered in stable nucleic acid lipid 364

particles (Geisbert et al., 2010) . Due to the greater length of long dsRNA when compared to their siRNA 365 counterparts, it is possible to have multiple viral genes sequences present in a single molecule. In theory, 366

this could induce knockdown of multiple viruses or multiple viral genes to inhibit infection, all without the 367 requirement of transfection or creation of multiple dsRNA fragments. When this was explored for the first 368 time in the present study, three combination molecules for the VSV N and M protein were shown to 369

significantly knockdown viral titers when cells were soaked with the long dsRNA. However, qRT-PCR 370

analysis revealed that only one of these molecules (5'N-3'M) was able to significantly reduce mRNA levels 371

of both viral genes. Though mRNA degradation is often associated with the knockdown observed during 372

RNAi The results of the present study indicate several unique findings. Firstly, this is the first time in 396 mammalian cells that RNAi has been observed through the natural uptake of sequence-specific dsRNA. 397

This indicates that it may be possible to develop antiviral therapies involving long dsRNAs that do not 398 involve costly transfection agents or stimulation of the damaging IFN response. Second, this viral inhibition 399 was observed to be length-dependent, as only dsRNA that was 300-400 bp in length or greater would induce 400

knockdown. This strongly implies a molecular channel such as SIDT2, although this was not explicitly 401 confirmed in our study. Finally, the success of combination dsRNA constructs suggest that it may be 402 possible to target either multiple genes within a single virus, genes originating from more than one virus or 403 possibly those from one virus along with associated host proteins. Moreover, we were able to provide 404 evidence that the observed viral inhibition was due to RNAi as Dicer1 knockouts could not induce this The the presence of cytopathic effects (CPE) and viral titers were calculated using the Reed and Meunsch 452 method to obtain the TCID50/mL (Reed & Muensch, 1938) . 453

Genes of interest were amplified using forward and reverse primers that contained T7 promoters. The 455

primer sets and their associated templates are outlined in Table 1 . The DNA products were amplified by 456 PCR using 10 ng of appropriate template (Table 1) , 2X GOTaq colorless master mix (Promega), 0.5 μM 457

of both forward and reverse primers ( Table 1 , Sigma Aldrich), and nuclease free water to a final volume 458 of 50 μL. The following protocol was carried out in a Bio-Rad T100 thermocycler: 98°C -5 min, 34 cycles 459 of 98°C -10s, 50°C -10s, 72°C -50s, followed by 72°C -5min. The resulting DNA amplicons with T7 460 promoters on both DNA strands were purified using a QIAquick PCR purification kit (Qiagen). The purified 461 product was then used in the MEGAScript RNAi Kit (Invitrogen) as per the manufacturer's instructions to 462 produce dsRNA. To confirm primer specificity, 100 ng of all PCR amplicons and the final dsRNA product 463

were separated on 1% agarose gels containing 1% GelGreen (Biotium Inc. In a 24-well plate, either THF or SNB75 were seeded at a density of 5.0 x 10 4 cells/well. Following 468 overnight adherence, the media was replaced before exposure to either a DPBS control, 0.5 μg/mL of long 469 dsRNA, 10 μg/mL of long dsRNA, or 10 μg/mL of high molecular weight (HMW) 470

polyinosinic:polycytidylic acid (pIC) all diluted in full growth media. Cells were exposed to these 471 treatments for 26h before the media was removed and the test wells were washed once with DPBS. Cells 472

were then collected in TRIzol (Invitrogen) and total RNA was extracted according to the manufacturer's 473 instructions. RNA was then treated with Turbo DNA-free™ Kit (Invitrogen) to remove any contaminating 474 genomic DNA. Complementary DNA (cDNA) was synthesized from 500 ng of purified RNA using the 475 iScript™ cDNA Synthesis Kit (Bio-Rad) following protocols provided by the manufacturer. of reverse primer (Sigma Aldrich) and nuclease-free water to a total volume of 10 µL (Fisher Scientific).

The sequences and accession number for each primer set are outlined in Table 2 . The qRT-PCR reactions 485

were performed using the CFX Connect Real-Time PCR Detection System (Bio-Rad). The program used 486 for all reactions was: 98°C denaturation for 3 min, followed by 40 cycles of 98°C for 5 sec, 55°C for 10 487 sec, and 95°C for 10 sec. A melting curve was completed from 65°C to 95°C with a read every 5 sec.

Product specificity was determined through single PCR melting peaks. All qRT-PCR data was analyzed 489

using the ΔΔCt method and is presented as the average of four experimental replicates with the standard 490 error of the mean (SEM). Specifically, gene expression was normalized to the housekeeping gene (β-491 actin) and presented as fold changes over the control group. well tissue culture plate and allowed to adhere overnight at 37°C with 5% CO2. All cell monolayers were 500

washed once with DPBS and then treated in eight-fold with a doubling dilution of dsRNA ranging from 501 800 ng/mL to 3.13 ng/mL for 24h at 37°C with 5% CO2 in normal growth media. Following incubation, 502 each well was washed twice with DPBS before exposure to AB and CFDA-AM as described previously by 503

Dayeh et al. (2003) . Because two fluorescent dyes were used to test cell viability, the 96-well plate was 504 read at an excitation of 530 nm and an emission of 590 nm for AB as well as an excitation of 485 nm and 505 an emission of 528 nm for CFDA-AM. The reads were completed using a Synergy HT plate reader (BioTek 506

Instruments). For each cell line analyzed, three independent experiments were performed. 507 508 4.6 Stimulating viral inhibition via soaking with low doses of dsRNA 509 THF and SNB75 cells were seeded at a density of 5.0 x 10 4 cells/well in a 24-well plate (Falcon). 510

Following overnight adherence, the media in all test wells was changed to fresh media. The cells were then 511 pre-treated for 2h with either a DPBS control, 500 ng/mL of 700 bp GFP dsRNA, or 500 ng/mL of 700 bp 512 mCherry dsRNA at 37°C with 5% CO2. Pre-treatment for 2h was selected after completing a time course 513 experiment to determine the optimal amount of time to pre-treat cells with dsRNA to induce viral 514 knockdown (supplementary figure S1 ). All test wells were then exposed to VSV-GFP at a multiplicity of 515 infection (MOI) of 0.1 and allowed to incubate for 24h at 37°C with 5% CO2 before supernatants were 516 collected for TCID50 quantification as described above. 517 518

To explore the impact that the dsRNA sequence length had on the observed viral knockdown, dsRNA 520

was synthesized to GFP that ranged in size from 200 bp to 700 bp and tested for ability to induce 521 knockdown. THF or SNB75 were seeded in a 24-well plate at a density of 5.0 x 10 4 cells/well. Following 522 overnight adherence followed by a media change, the cells were pre-treated for 2h with either a DPBS 523 control, 500 ng/mL of mCherry dsRNA, or 500 ng/mL of GFP dsRNA at lengths of 200 bp, 300 bp, 400 524 bp, 500 bp, 600 bp and 700 bp at 37°C with 5% CO2. All test wells were then exposed to VSV-GFP at an 525 MOI of 0.1 and allowed to incubate for 24h at 37°C with 5% CO2 before supernatants were collected for 526 TCID50 quantification as described above. 527 528

4.8 Use of viral genes for dsRNAi 529 4.8.1 VSV-GFP 530

DsRNA was synthesized to the VSV viral genes of N protein and M protein as described above. 531

Either THF or SNB75 were seeded in a 24-well plate at a density of 5.0 x 10 4 cells/well. Following 532 overnight adherence, the media in all test wells was changed to fresh media. The cells were then pre-533 treated for 2h with either a DPBS control, 500 ng/mL of VSV N protein dsRNA, 500 ng/mL of VSV M 534 protein dsRNA, 500 ng/mL of mCherry dsRNA or a combination of 250 ng/mL of VSV N protein and 535

250 ng/mL of VSV M protein (500 ng/mL total of dsRNA) at 37°C with 5% CO2. All test wells were then 536 exposed to VSV-GFP at an MOI of 0.1 and allowed to incubate for 24h at 37°C with 5% CO2 before 537 supernatants were collected for TCID50 quantification as described above. 538

DsRNA was synthesized for HCoV-229E viral genes of RdRp, Spike protein, N protein and M 540 protein as described above. MRC5 cells were seeded in a 24-well plate at a density of 7.5 x 10 4 cells/well. 541

Following overnight adherence, the media in all test wells was changed to fresh media. The cells were 542 then pre-treated for 2h with either a DPBS control, 500 ng/mL of 229E RdRp, 500 ng/mL of 229E Spike 543 protein, 500 ng/mL of 229E N protein dsRNA, 500 ng/mL of 229E M protein dsRNA or 500 ng/mL of 544 mCherry dsRNA at 37°C with 5% CO2. All test wells were then exposed to HCoV-229E at an MOI of 545 0.02 and allowed to incubate for 24h at 37°C with 5% CO2 before supernatants were collected for TCID50 546 quantification as described above. 547

DsRNA was synthesized for the SARS-CoV-2 viral genes, N protein and M protein, as described 549

above. Calu-3 cells were seeded in a 12-well plate at a density of 2.0 x 10 5 cells/well. Two days later, the 550 media was replaced with fresh media. The cells were then pretreated for 2h with either 1000 ng/mL of 551 mCherry dsRNA control, 1000 ng/mL of SARS-CoV-2 M protein dsRNA or 1000 ng/mL of SARS-CoV-552 2 N protein dsRNA at 37°C with 5% CO2. Following pre-treatment, the cells were exposed to SARS-553

CoV-2 at an MOI of 1.0 for 1h, washed twice with sterile 1x PBS, and the dsRNA added back to the 554 appropriate wells. After 24h, total RNA isolation was performed using the RNeasy Mini Kit (Qiagen) 555

according to the manufacturer's protocol. SARS-CoV-2 specific genome levels were measured by qPCR 556

using for the basolateral side. The DPBS was removed from the apical side of the test transwells and were then 578 exposed to either media alone, 500 ng/mL of dsRNA (VSV N protein, HCoV-229E M protein or mCherry 579 as a control) or 50 μg/mL of pIC for 2h at 37°C with 5% CO2. Following the 2h incubation, appropriate 580 test wells were exposed to either VSV-GFP (MOI = 0.1) or HCoV-229E (MOI = 0.1) and incubated for 581

24h before the supernatants were collected and the TCID50 was quantified as described above. 582 583

4.10 Soaking versus transfection with siRNA 584

In order to directly compare the effects of soaking with long dsRNA or siRNA on virus inhibition, 585

THF and SNB75 cells were seeded in 24-well plates at a density of 5.0 x 10 4 cells/well. Following overnight 586

adherence and a fresh media change, cells were exposed to either a DPBS control, 2 nM of 700 bp GFP 587 dsRNA, 2 nM of GFP Silencer ® siRNA (Ambion) or 2 nM of the negative control Silencer ® siRNA 588 (Ambion) for 2h at 37°C with 5% CO2. Cells were exposed to nanomolar concentrations (equivalent to 500 589 ng/mL of 700 bp dsRNA) to ensure that the same number of dsRNA and siRNA molecules were added in 590 each treatment group. Following this incubation, wells were exposed to VSV-GFP at an MOI of 0.1 and 591

incubated for 24h at 37°C with 5% CO2 before supernatants were collected for TCID50 quantification as 592 described above. 593

For validation that the siRNA molecules were functional and capable of inducing knockdown, the 594 siRNA molecules were transfected into SNB75 and THF cells and subsequent viral numbers were 595

quantified. THF and SNB75 cells were seeded 24-well plates at a density of 5.0 x 10 4 cells/well. Following 596 overnight adherence and a fresh media change, cells were 10 nM of GFP Silencer ® siRNA (Ambion) or 10 597 nM of the negative control Silencer ® siRNA (Ambion) was transfected into the cells using Lipofectamine 598

RNAiMAX (Invitrogen). Cells were transfected with 10nM siRNA as recommended by the manufacturer. 599

Following a 24h incubation at 37°C with 5% CO2, wells were washed twice with DPBS and then exposed 600 to VSV-GFP at an MOI of 0.1 and incubated for 24h at 37°C with 5% CO2 before supernatants were 601 collected and the TCID50 was quantified as described above. 602 603 4.11 Inducing viral inhibition using combination dsRNA molecules that target multiple viral genes 604

To determine whether combination dsRNA could induce viral knockdown via inhibition of multiple 605 viral genes at once, THF cells were seeded at a density of 5.0 x 10 4 cells/well in a 24-well plate. Following 606 overnight adherence, the media in all test wells was changed to fresh media. The cells were then pre-treated for 700 bp), or 500 ng/mL mCherry dsRNA at 37°C with 5% CO2. All test wells were then exposed to 611

VSV-GFP at an MOI of 0.1 and allowed to incubate for 24h at 37°C with 5% CO2 before supernatants were 612 collected for TCID50 quantification as described above. The cell monolayers were collected in Trizol so 613 that total RNA could be extracted and cDNA was synthesized as described above in section 2.4. The 614

expression of VSV genes (M and N protein) was measured by qRT-PCR using the same method as outlined 615 above in section 2.4. The sequences and accession number for the primer sets used here are outlined in 616 Table 2 . The VSV gene expression of cells exposed to the 5'M-3'N molecule was not measured due to the 617 small size of the M protein gene which made it impossible to develop qPCR primers that did not amplify a 618 region of the M-N dsRNA that was used to soak the cells. 619 620 4.12 Dicer knockout studies 621

For successful knockdown, the RNAi pathway requires the use of Dicer to cleave viral RNAs into 622

siRNAs. To provide evidence that the knockdown observed here was due to RNAi, a Dicer1 knockout 623 mouse MSC cell line (Dicer1 -/-) was used along with its corresponding functional Dicer1 cell line (Dicer1 624 f/f). For each experiment, both the knockout and functional Dicer MSC cell lines were seeded at a density 625 of 5.0 x 10 4 cells/well in a 24-well plate. Following overnight adherence, the media in all test wells was 626 changed to fresh media. Both cell types were then pre-soaked for 2h with either a DPBS control, 500 ng/mL 627 of VSV N protein dsRNA, or 500 ng/mL of mCherry dsRNA at 37°C with 5% CO2. All test wells were Table 1 : Primers with underlined T7 promoter sequences that were used for amplification of genes of

Long dsRNA induces 710 an antiviral response independent of IRF3, IPS-1 and IFN

DsRNA and the innate antiviral immune response

Sensitivities of human glioma cell lines to interferons and double-stranded 715

RNAs individually and in synergistic combinations

Duplexes of 21-nucleotide 717 RNAs mediate RNA interference in cultured mammalian cells

TLR-driven early 719 glycolytic reprogramming via the kinases TBK1-IKKε supports the anabolic demands of dendritic cell 720 activation

Transport of dsRNA into cells by the transmembrane protein

Production of antisense RNA leads to effective 724 and specific gene expression in C. elegans muscle

Potent and specific genetic 726 interference by double-stranded RNA in Caenorhabditis elegans

Effects of interferons and viruses on metabolism

RNA interference of 730 influenza virus production by directly targeting mRNA for degradation and indirectly inhibiting all 731 viral RNA transcription

Postexposure 733 protection of non-human primates against a lethal Ebola virus challengs with RNA interference: a 734 proof-of-concept study

Short interfering RNA confers intracellular antiviral immunity 736 in human cells

par-1, a gene required for establishing polarity in C. elegans embryos, 738 encodes a putative Ser-Thr kinase that is asymmetrically distributed

Double-stranded RNA activates type I interferon 740 secretion in glomerular endothelial cells via retinoic acid-inducible gene (RIG)-1

Length-dependent accumulation of double-743 stranded RNAs in plastids affects RNA interference efficiency in the Colorado potato beetle

The roles of two IκΒ 746 kinase-related kinases in lipopolysaccharide and double stranded RNA signaling and viral infection

Fibroblasts: the unknown sentinels eliciting immune responses against microorganisms

ISG54 and ISG56 are 752 induced by TLR3 signaling in U373MG human astrocytoma cells: possible involvement in CXCL10 753 expression

Sensing of RNA viruses: a review of innate immune receptors involved 755 in recognizing RNA virus invasion

Length-dependent 757 recognition of double-stranded ribonucleic acids by retinoic acid-inducible gene-I and melanoma 758 differentiation-associated gene 5

Dicer functions in 760 RNA interference and in synthesis of small RNA

TLR3 agonists: RGC100, ARNAX and poly-IC: a comparative 763 review. Imunologic Research

Intracellular sensing of viral genomes and viral evasion

The TLR3 767 signaling complex forms by cooperative receptor dimerization

Type I interferons: distinct biological 769 activities and current applications for viral infection

Systemic RNA interference deficiency-1 (SID-1) 772 extracellular domain selectively binds long double-stranded RNA and is required for RNA transport by 773 SID-1

Small interfering RNA-mediated 775 translation alters ribosome sensitivity to inhibition by cycloheximide in Chlamydomonas reinhardtii

Dicer function 778 as an antiviral system against human adenoviruses via cleavage of adenovirus-encoded noncoding 779 RNA

Antiviral RNA interference 781 in mammalian cells

Inactivation of the type I interferon pathway reveals long double-stranded RNA-mediated RNA 784 interference in mammalian cells

Matskevich AA, Moelling K (2007) Dicer is involved in protection against influenza A virus infection

Assessment of the toll-like receptor 790 3 pathway in endosomal signaling

Evaluation 792 of single and dual siRNAs targeting rabies virus glycoprotein and nucleoprotein genes for inhibition of 793 virus multiplication in vitro

Extracellular dsRNA: its function and mechanism of cellular uptake

SIDT2 transports 797 extracellular dsRNA into the cytoplasm for innate immune recognition

Stable suppression of gene expression by RNAi in 799 mammalian cells

Functional anatomy of a dsRNA trigger: differential 801 requirement for the two trigger strands in RNA interference

Understanding viral dsRNA-mediated innate immune responses at 803 the cellular level using a rainbow trout model

Flavivirus induces and antagonizes 805 antiviral RNA interference in both mammals and mosquitoes

A simple method of estimating fifty per cent endpoints

RIG-I-like receptors: their regulation and roles in RNA sensing

A new two-stranded helical structure: polyadenylic acid and polyuridylic 812 acid

A role for human 814 Dicer in pre-RISC loading of siRNAs

The endocytic pathway 816 mediates cell entry of dsRNA to induce RNAi silencing

Inosine-containing dsRNA binds a stress-granule-like complex and downreguates 818 gene expression in trans

Antiviral RNAi in insects and mammals: parallels and 820 differences

Reciprocal inhibition 822 between intracellular antiviral signaling and the RNAi machinery in mammalian cells

Protection 825 of mice against lethal rabies virus challenge using short interfering RNAs (siRNAs) delivered through 826 lentiviral vector

Double-stranded RNA is detected by immunofluorescence 828 analysis in RNA and DNA virus infections, including those by negative-stranded RNA viruses

Extracellular RNA sensing by pattern 831 recognition receptors

Double-833 stranded RNA is internalized by scavenger receptor-mediated endocytosis in Drosophila S2 cells

RISC-target interaction: cleavage and translational suppression

The 838 RIG-I-like receptor LGP2 inhibits Dicer-dependent processing of long double-stranded RNA and 839 blocks RNA interference in mammalian cells

Hepatitis C virus core 841 protein is a potent inhibitor of RNA silencing-based antiviral response

Molecular Structure of Nucleic Acids: A Structure for 843

Double-stranded RNA is 845 produced by positive-strand RNA viruses and DNA viruses but not in detectable amounts by negative-846 sense RNA viruses

Durable 848 knockdown and protection from HIV transmission in humanized mice treated with gel-formulated CD4 849 aptamer-siRNA chimeras

Systemic RNAi in C. elegans requires the putative 851 transmembrane protein SID-1

Variable deficiencies in the interferon response 853 enhance susceptibility to vesicular stomatitis virus oncolytic actions in glioblastoma cells but not in 854 normal human glial cells

Developing effective siRNAs to reduce the expression of key viral genes of 856 COVID-19

All about the RNA: interferon-stimulated genes that interfere with viral RNA 858 processes

RNA interference functions as an antiviral immunity 860 mechanism in mammals

Specific double-stranded RNA interference in 862 undifferentiated mouse embryonic stem cells

Human Dicer preferentially cleaves 864 dsRNAs at their termini without a requirement for ATP

then exposed to VSV-GFP at an MOI of 0.1 and allowed to incubate for 24h at 37°C with 5% CO2 before 629 supernatants were collected for TCID50 quantification as described above. 630 6314.13 Statistical analyses 632All data sets were tested for a normal distribution (Shapiro-Wilk) and homogeneity of variance 633 (Levene's) using R and RStudio (R Core Team, 2014; RStudio Team, 2015). Further statistical analyses 634were also completed using R and RStudio. For the viability, VSV gene expression and viral titer data, a 635one-way analysis of variance (ANOVA) was completed followed by a Tukey's post-hoc test to compare 636 between all exposure conditions. When determining whether the IFN genes were upregulated, a one-way 637 ANOVA was completed followed by a Dunnett's multiple comparisons post-hoc test to detect significant 638 differences from the control condition. With the siRNA transfection data, a two-tailed unpaired t-test was 639completed. For all statistical analyses, a p-value less than 0.05 was considered significant. All data is 640presented as the average of experimental replicates + SEM.

Acknowledgements: 643The authors would like to acknowledge Dr. Tamiru 10 μg/mL as well as with HMW pIC at a concentration of 10 μg/mL. Following treatment, transcript 929 expression of IFNβ (i) and CXCL10 (ii) was assessed via qRT-PCR analysis. All data were normalized to 930 the reference gene (β-Actin) and expressed as a fold change over the control group where control expression 931 was set to 1. Error bars represent +SEM, and represents the average of 3 independent replicates. A p-value 932 of less than 0.001 is represented by a *** symbol while a p-value of less than 0.0001 is represented by a 933 **** symbol when compared only to the control (Ctl) treatment. 934 and MRC5 (iii) were soaked with 700 bp dsRNA for 26h at concentrations that ranged from 0 ng/mL to 936 800 ng/mL. Cellular metabolism was measured using an Alamar Blue assay (A) and membrane integrity 937 was measured using CFDA (B). Error bars represent +SEM, and each data point represents the average of 9383 independent experiments. A p-value of less than 0.05 was considered to be statistically significant. Error 939 bars with different letters represent significantly different data. 940 Appearance of the THF cells after treatments with dsRNA and VSV-GFP infection as observed under the 945 fluorescent microscope at 50X magnification (C). Error bars represent +SEM, and each data point 946represents the average of 6 independent replicates. A p-value of less than 0.05 was considered to be 947 statistically significant. Error bars with different letters represent significantly different data. 948 HCoV-229E sequences for either RdRp, M protein, N protein and the spike protein before 24h infection 955with HCoV-229E (MOI = 0.02) (C). Calu-3 cells were pre-soaked for 2h with either DPBS alone, 1000 956 ng/mL of the mCherry mis-matched dsRNA sequence control or 1000 ng/mL of dsRNA matching SARS-957CoV-2 sequences for either M protein and N protein prior to 24h infection with SARS-CoV-2 (MOI = 1.0) 958 (D). Error bars represent +SEM, and each data point represents the average of 6 independent replicates. A 959 p-value of less than 0.05 was considered to be statistically significant and different letters represent 960 significant differences. For the SARS-CoV-2 data, a p-value of less than 0.01 is represented by a ** symbol 961 and a p-value of less than 0.05 is represented by a * symbol when compared only to the control treatment 962 The pBECs were also pre-treated with either DPBS, 50 μg/mL of HMW pIC, 500 ng/mL of the mis-matched 968 mCherry dsRNA control or 500 ng/mL of HCoV-229E M protein dsRNA before infection with HCoV-969 229E (MOI = 0.1) for 24h (C). Error bars represent +SEM, and each data point represents the average of 3 970 independent replicates. A p-value of less than 0.05 was considered to be statistically significant. Error bars 971 with different letters represent significantly different data. 972 Error bars represent +SEM, and each data point represents the average of 5 independent replicates. A p-978 value of less than 0.05 was considered to be statistically significant and different letters represent significant 979differences. For the transfection data, a p-value of less than 0.01 is represented by a ** symbol while less 980 than 0.001 is represented by a *** symbol. 981 with DPBS or 1000 ng/mL of either mCherry, 5'N-3'M, 5'M-3'N or N-M Alt before being exposed to 988 VSV-GFP (MOI = 0.1) for 24h (Ci). Following this treatment, cells were collected and RNA extracted so 989 that gene expression of the VSV N protein gene (Cii) and M protein gene (Ciii) could be measured by qRT-990PCR. Error bars represent +SEM. Each data point for the titer data represents the average of 6 independent 991replicates while the qRT-PCR data represents the average of 5 independent replicates. A p-value of less 992 than 0.05 was considered to be statistically significant. Error bars with different letters represent 993 significantly different data. 994 M14 Cells (75,000 cells/well) were exposed to 500 ng/mL of each dsRNA (700 bp each) at various times 1072 before infection with VSV-GFP (MOI = 1). Following 24 hours of infection, supernatants were collected 1073 and the TCID50 was calculated using HEL-299 cells. This has been repeated three times. Significant 1074 differences were assessed between mCherry and GFP at each individual timepoint using a Sidak's 1075 multiple comparisons test.